The Text Analytics Market(s)

The Text Analytics Market(s)

Where semantically
-
aware technologies play a role


by

Curt A. Monash, Ph.D.

President, Monash Research

Editor,
Text Technologies

contact@monash.com

http://www.monash.com

http://www.texttechnologies.com

8 linguistics
-
based businesses


Web search


Public
-
facing site search


Enterprise search & knowledge management


Custom publishing


Text mining and extraction


Spam filtering


Voice recognition and NLP


Machine translation

Web search



Giant oligopoly


Startups aren't gaining traction …


… and neither are leaders’ whizz
-
bang features


Three kinds of search


Navigational


Innovation = Maps, part
-
of
-
page smartness


Informational


Innovation is scattershot …


… because people are content with how this works


Transactional


Big bucks


Big adversarial information retrieval problem


Semantics isn’t the issue




Web search vendor issues today



Physical efficiency


Adversarial information retrieval


Monetization


Regulation


Branding …


… none of which are about semantics


Future web search issues



Better popularity metrics


Better use of context (tagging helps)


Geography/geocoding!!


Sub
-
page retrieval (tagging will be huge)


Looking through security barriers


Better UIs (Clusters/facets? Voice?)


Finally a role for sophisticated semantics!


But how good does the technology have to get before
people care?



One
-
size
-
fits
-
all enterprise search failed


Overreaching had a lot to do with that



E
-
commerce search isn't general search


Text mining isn't search (or clustering)


Custom publishing isn't exactly search


Public
-
facing site search



It has dual goals


What the user actually wants


What you want the user to see


E
-
commerce and general search are separate
businesses


E
-
commerce = province of smaller vendors


General = province of search, portal, or CMS vendors


Hand
-
tagging is key …


… or else generic Google is usually superior


Role of sophisticated semantics starts and usually
ends w/ hand tagging


Many reasons to buy true enterprise search



General productivity


Sure beats looking through file cabinets


“Wouldn't it be nice if ...”


KM dreams


Compliance


“Thou shalt …


Search specialized external databases


Medline


News/National intelligence


Taxonomies and RDF both play well here


Bollixed enterprise search market landscape



Multiple generations of vendors flamed out


Verity


Excalibur (endlessly backed by Allen & Company)


Fulcrum et al.


FAST


Market dynamics are screwy


Google has brand name, ease of installation


Microsoft, Oracle, SAP, and Autonomy are all confused


Small vendors are ... small


Products are sold for inappropriate apps


Compliance is driving demand


Facets and clusters are popular





Kinds of enterprise search



Find a specific document


Find an answer


Find ALL documents


Find an expert


Semantic technologies could help with all of those


Technical challenges in enterprise search



Documents may not exist


Many formats


Corpus weighting


No good popularity metrics


Security


Ease of adoption


The problem isn’t
just

insufficiently good semantics

Custom publishing



More precise than search


Sophisticated extraction


Focus on document
parts


Started with technical publishers (and intelligence)


Expanded to


Other technical documents


Other publishers


Semantics is central to this market


This is one of the top application areas for extraction


Text mining



Really about sophisticated extraction


Apps and verticals mirror data mining


The action is in sentiment analysis ...


... and ease of use


Otherwise the industry seems tired



The original text mining apps: Early warning

Source: July, 2006 post on
Text Technologies



Vehicle safety


Other manufacturing/warranty analysis apps


Reputation management


Other customer sentiment apps


Anti
-
terrorism


Sarbanes
-
Oxley compliance


(continued …)

More early warning

(… continued)



Antifraud


Stopping money laundering


Clinical applications (some)


Early insurance risk management apps


Early experimental hedge fund apps


Employee (dis)satisfaction (missing from the
original list)


Customer/market intelligence


Now drive text mining growth


Internal Voice of the Customer came first


Voice of the Market has blended in


Sales, buying, and delivery practices are in line


Departmental buying


SaaS delivery


Many vendors focus(ing) on this segment:


Attensity


Clarabridge


SAS


SPSS


Expert System S.p.A



Three ways to use text mining/extraction


Business intelligence


Reports, dashboards


Attensity, Clarabridge


Predictive analytics


Data mining


SAS, SPSS


Custom publishing


nStein et al.


Lots of partnerships


Linguistics all the way down

In this market, semantics
is

central



Sophisticated taxonomies


Sentiment extraction!


Syntax is often important too

Six trends that could shake up the market



Web/enterprise/messaging integration


BI integration


Universal message retention


Portable personal profiles


Electronic health records


Voice command & control …


… and semantics plays (or should play) a big role in
all of them.




Further information


Curt A. Monash, Ph.D.

President, Monash Research

Editor,
Text Technologies


contact@monash.com

http://www.monash.com

http://www.texttechnologies.com